AbstentionBench - underspecified context - GPQA-Diamond - Precision: leaderboard
Metric: Abstention Precision (%). Source: github.com. 21 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Llama 3.1 8B Instruct | 100 |
| 2 | GPT-4o | 100 |
| 3 | Llama 3.1 405B Instruct | 100 |
| 4 | Qwen 2.5 32B Instruct | 100 |
| 5 | O1 | 100 |
| 6 | Llama-3.1-Tulu-3-8B | 100 |
| 7 | Llama-3.1-Tulu-3-8B-DPO | 100 |
| 8 | O1 (Low) | 100 |
| 9 | Llama-3.1-Tulu-3-70B | 96.3 |
| 10 | Llama-3.1-Tulu-3-70B-DPO | 96.3 |
| 11 | Gemini 1.5 Pro | 95.24 |
| 12 | Llama 3.1 70B Instruct | 94.74 |
| 13 | Llama 3.3 70B Instruct | 93.75 |
| 14 | O1 (High) | 92.31 |
| 15 | Llama 3.1 70B | 90 |
Interactive version: theaggregate.ai/benchmark?slug=abstentionbench-underspecified-context-gpqa-diamond-precision · How It Works · Data refreshed daily, snapshot 2026-09-19.